Perhaps the most interesting conversation I hope to have at BarCamp this weekend is on these thoughts on the social graph problem, by Brad Fitzpatrick with contribution by David Recordon. Dave, btw, is stuffing bottles for BarCamp as we speak and I just met him for the first time today. So I might go add him as a on Facebook or somewhere and maybe as a friend a whole bunch of places later. You know the drill.
A long time ago we talked about standards for social networking. We had Marc Canter beating the FOAF bongo and all knew when Social Networking needed openess before it closed. Not just when we moved en mass from Ryze to Friendster to Tribe to Orkut and the 500+ YASNS that followed, but started straddling social tools. There is no doubt about the user need for portability with identity and relationships (although faceted), and significant commercial need for building networks and marketing to them. Such incentives concern me.
I've been trying to think through ways the proposed system could be gamed. Basically, so long as you still have to opt into each graph, control the facet of your identity you are expressing, and opt-in to each relationship in each context -- most of the privacy concerns could be controlled. There are security threats that need many eyes on the problem. A standard to code towards and a clearinghouse for data will undoubtedly unleash constructive innovation.
Perhaps it is the apps that ride the graph themselves that are the concern. You probably already understand how an app, lets say through Facebook, can be designed for extreme virality with little social consideration. But how an app traverses an opted-in graph is of less concern as each SNS or graph has incentives to limit this and the proposed standard limits scope to identity and relationships, not profile data.
The one concern I can come up with, and it could be addressed, is the power given to third party developers. If you are a developer as described, you could find out what graphs a given identity is a part of. It isn't clear if a developer gets access to relationships post-opt-in towards the developer on a per user basis.
Like I said, I'm looking forward to the conversation. The concept is fascinating and it is better to have more eyes on the opportunity.
It's those incentives that scare me, along with new potential for relationship theft. It may just be part of the world that's coming, but consider how it could play out. A Fakester or two attracts you and your friends. Then joins another social network that has a completely different social context. It isn't just if you opt into the graph. It may not even matter entirely because enough edges are around you.